Assumption mapping: how to run the session and pick your first test

Assumption mapping is a team exercise where you write down everything that must be true for an idea to work, plot each belief by how important it is and how much evidence you have, and test the important ones with the least evidence first. The best-known version comes from David J. Bland of Precoil and the book Testing Business Ideas, and it sorts assumptions into desirability, viability and feasibility. This post covers how to run one session (we suggest blocking about two hours) and how to turn the top-right corner of the map into a first test you can start this week.

The map itself is easy. The hard parts are writing assumptions specific enough to be wrong, and refusing to let the loudest person in the room decide what counts as evidence.

1. Know what the exercise is for before you book the room

Assumption mapping exists because teams often run familiar experiments, such as a landing page or a round of interviews, and ignore the beliefs that could actually sink the idea. Bland says this directly in his Strategyzer excerpt from Testing Business Ideas, published August 4, 2020: the map is meant to point testing effort at the riskiest beliefs.

The 2x2 is older than Bland's version. Precoil says he learned it from Jeff Gothelf, Josh Seiden and Giff Constable, saw that teams still struggled with what type of experiment to run, and added the desirable, viable and feasible themes from design thinking as colors. The same page notes the method was included in Google's Design Sprint Kit and is published under Creative Commons, so you can use the worksheet freely.

Use it for a new product, a new market or a new business model, where most of what you believe is unproven. It fits poorly on a mature product making small changes. Paweł Huryn's critique on The Product Compass argues it suits new products better than continuous discovery, and that is a fair line to draw.

2. Sort assumptions into the three categories, and add a fourth if the market is moving

Each category is a question. Bland's Strategyzer excerpt frames each one as the question below. The examples are ours.

Adaptability is the fourth type in the Strategyzer excerpt. The Precoil and Mural versions stop at three, and for a first session three is plenty. Add adaptability when your idea leans on something outside your control, such as a platform's rules or a pending regulation.

Use one sticky note color per category. The Mural template guide by Annelise Schoups uses orange for desirability, blue for feasibility and green for viability. The colors matter more than they look: a map that is almost all blue tells you the team has spent its time worrying about the build.

3. Write each assumption so it could turn out to be false

Start the session with one question: what must be true for this idea to work? Bland recommends phrasing each answer as "We believe that..." so it reads as something that could be wrong, and he asks that each one be testable, precise and about one thing (Strategyzer). The Mural guide adds one statement per note, no bullet points, and a ground rule to base everything on what you know today.

Here is a made-up example that runs through the rest of this post. Dana and two colleagues are building software that lets independent veterinary clinics send pet owners refill reminders for prescription medication. Their first batch of notes looked like this:

• "Vets will love it." This is too vague to test. Who, love what, shown how?
• "We believe that practice managers at independent clinics lose refill revenue because owners forget to reorder." This is testable.
• "We believe that clinics will pay $150 a month." This is testable, and viability.
• "We believe that we can read prescription data from the two most common practice systems." This is testable, and feasibility.

Rewrite the vague ones on the spot. If nobody can say what result would prove a note wrong, it is a mood, and moods do not go on the map.

As a rough target, aim for 15 to 30 notes. Fewer usually means the team skipped a category, and more usually means people are writing the same belief three ways.

4. Place notes on two axes, one category at a time

Draw a 2x2. In Bland's version the vertical axis is importance (important at the top, unimportant at the bottom) and the horizontal axis is evidence (have evidence on the left, no evidence on the right), per the Strategyzer excerpt. To judge importance, ask which beliefs would make the idea fail if they were wrong.

Map the desirability notes first, then feasibility, then viability. Bland gives that order in the same excerpt (with adaptability last, if you use it). The reason to start there: if customers do not want it, whether you can build it or make money from it stops mattering. The unsigned Strategyzer book summary makes a similar warning against spending on technology before the customer problem is confirmed. Placing one color at a time also stops the session from turning into a debate about the database by minute ten.

Place notes relative to each other. Put the first one in the middle, then ask of each new note: more or less important than that one, and more or less evidence? Relative placement goes faster than arguing over absolute scores, and the order is what you need.

Short version: one category at a time, compare each note to its neighbors, and only importance and evidence decide where it goes.

5. Hold a strict line on what counts as evidence

Evidence is data, and Bland says it can be qualitative or quantitative. He is blunt that who argues loudest or who is paid the most does not decide it (Strategyzer). Interview notes count. A strong hunch from the most senior person in the room does not.

When someone says "we know this," ask one follow-up: how do we know? In the made-up example, Dana's cofounder was sure owners forget refills because his own dog's medication ran out twice last year. That is a real experience, and it is one owner, and the note goes on the right side of the map.

The Strategyzer book summary of Testing Business Ideas, which carries no author's name, ranks evidence strength this way: actual behavior such as purchases or usage is strong, stated intentions and preferences are moderate, and opinions and hypothetical answers are weak. Use that ranking to place notes. A note backed only by people saying they would buy sits further right than one backed by people who did buy.

6. Pick the first test from the top-right corner, and only one note from it

The top-right quadrant holds the beliefs that matter most and have the least behind them, and Bland says that is where near-term experiments should go (Strategyzer). The Mural guide adds the consequence: if one of these turns out false, plans built on it may need rethinking.

The corner will usually hold several notes. Choose the one that, if wrong, would make the others irrelevant. In the made-up example, the top-right corner held three notes: practice managers lose refill revenue to forgetful owners, clinics will pay $150 a month, and the prescription data can be read. If practice managers do not see lost refills as a problem, the price and the integration stop mattering. Test that one first.

Leave the bottom half alone. The Strategyzer excerpt advises against spending effort on beliefs that are unimportant and already supported, and the same logic applies to the unimportant ones you have no evidence for: they are cheap to be wrong about.

Huryn's suggestion to run one or two experiments at a time, learn, and repeat (The Product Compass) is a good cap for a small team. More than two in parallel and nobody remembers which result belongs to which note.

7. Write the test down before you run it, with the pass line set in advance

Use a test card. Alex Osterwalder introduced it on Strategyzer in March 2015 as four questions: what must be true, how you will test it, what you will measure, and what result counts as success, including the threshold. Filled in for the made-up example:

• We believe that practice managers at independent clinics lose refill revenue because owners forget to reorder.
• To verify that, we will hold 10 short calls with practice managers at independent clinics and ask how refills worked the last time one was late.
• We will measure how many can describe a specific recent lapse and whether they currently do anything to chase refills.
• We are right if at least 6 of the 10 describe a lapse from the past month and already spend staff time chasing it.

Set the pass line before the first call. Once results are in, every team discovers that 4 out of 10 was the number they meant all along. (It was not.)

Start with the cheapest test that could prove the note wrong. The Strategyzer book summary says the book describes 44 experiment types compared by setup time, run time, evidence strength and cost, and recommends starting with the smallest test that can confirm or refute an assumption. The summary's rough figures make the case: it says customer interviews take 1 to 2 days to prepare and about 30 minutes each, while a minimum viable product takes 2 to 8 weeks to set up and 4 to 12 weeks to gather usage data.

Interviews are the right first test for a desirability note like Dana's. They are weak on willingness to pay, because what people say they would spend is a stated intention. When the price note reaches the top of the list, move to something that asks for a commitment, such as a fake door test or a paid pilot, as covered in market validation.

8. Move the notes after every test, and rerun the map when the corner empties

After each test, move the note. A confirmed belief slides left toward evidence. A refuted one either comes off the board or changes the idea, and Huryn notes that a test can also shift how important a belief looks (The Product Compass).

In the made-up example, if only 2 of 10 practice managers had described a recent lapse, Dana's team would not move on to the price test. They would ask what those 10 managers do worry about, which is the job of product discovery and interviews built on The Mom Test.

Keep the map on a wall or in a shared board and revisit it every week or two while the idea is young. A map that was made once for a workshop and never touched again is a nice photo for the slide deck.

9. Know the common objections, and where they are right

Huryn's critique raises two fair points about Bland's version (The Product Compass). "Does the market want this?" is broad enough to hide separate risks: whether the product is valuable, whether people can use it, and whether you can reach buyers. And "importance" can mean different things to different people in the room.

Fix both in the session itself. Split desirability notes into separate value, usability and reach notes when they blur together. Define importance once, at the start, as "the idea fails if this is wrong," and hold everyone to it.

Who it does not fit: a team adding a feature to a product with steady usage data usually learns more from that data than from sticky notes. A solo founder can do it alone, though the method's main value is surfacing disagreements, and you will mostly agree with yourself.

Your first step

Book two hours with the people who would build, sell and fund the idea, ask what must be true for it to work, and leave with one test card for one top-right note. If that first test needs conversations with specific people you cannot easily reach, such as practice managers at independent vet clinics, Instant Expert can find people who match the description and send them paid invitations for a short call or a written answer, and you pay only if someone books or answers. For help running the session, see the directory pages for product management consultants and startup advisors. The value proposition canvas is a good source of desirability notes before you start.